Graphing the offered load or rate of a given flow is very useful for reporting
and visualizing the behavior of an application. Argus provides a great deal of flexibility
in reporting and graphing flow data, but the data can also be used by other programs,
like Excel and Mathmatica.

In this series of graphs, we are plotting the average bandwidth reported in each of the
5 second status reports that argus generated for a specific iTunes based Internet radio
stream, that last for an entire day. We are plotting the data as a histogram, or a
frequency distribution, so you can see the variation of the demand over the lifetime of
stream. This is extremely useful when characterizing the network resources needed to support
a given application, but it also gives a great indication of the performance of the network
during the day. For applications like video and voice/sound distribution, the streams
should maintain a pretty constant load, and watching the changes can give some insight into
the customers quality of experience, without having to ask the user.

The first graph is what the application looks like on a good day. The second graph
is the same radio channel, same workstation, but on a bad day for easy listening.
The primary difference between the two graphs, is a broadening of the variance in
measured bandwidth, which is indicative of periods of loss and the subsequent retransmission
needed to suppport TCP's reliable data transport.

The graph was generated using argus-3.0, rahisto(), Excel and Mac OS X:

  rahisto -r datafile -H drate 140:100-170K